Formation of High-Quality GaN Microcrystals by Pendeoepitaxial

COMMUNICATION pubs.acs.org/crystal

Formation of High-Quality GaN Microcrystals by Pendeoepitaxial Overgrowth of GaN Nanowires on Si(111) by Molecular Beam Epitaxy Pinar Dogan,* Oliver Brandt, Carsten Pf€uller, Jonas L€ahnemann, Uwe Jahn, Claudia Roder, Achim Trampert, Lutz Geelhaar, and Henning Riechert Paul-Drude-Institut f€ur Festk€orperelektronik, Hausvogteiplatz 5 7, 10117 Berlin, Germany ABSTRACT: Hexagonal GaN microcrystals of a size between 1 to 3 μm are obtained by the pendeoepitaxial overgrowth of a GaN nanowire template on Si(111). The GaN microcrystals are free of threading dislocations and exhibit an atomically smooth surface (roughness of 0.2 nm). Photoluminescence spectra of these microcrystals are dominated by an intense donor-bound exciton transition at 3.471 eV with a width of 1 meV reflecting strain-free GaN of exceptional structural quality.

G

aN and the related Al and In containing group-III nitride alloys have established themselves as the material of choice for the fabrication of light-emitting and laser diodes in the visible spectral range. However, to fully exploit the potential of these devices, their miniaturization is required. Two-dimensionally positioned microsized light-emitting and laser diodes will enable, for example, high-density light-emitting arrays with individually controllable pixels and ultralow threshold, respectively. For the realization of economically viable solutions, the integration of these microdevices on Si substrates is of particular interest. Previous work toward the fabrication of these microdevices followed a top-down approach involving lithography and subsequent processing of as-grown GaN films. For instance, optically pumped GaN microdisk lasers operating at room temperature have been demonstrated by using wet chemical etching of (Al,In)N sacrificial layers.1 Photoelectrochemically defined (In,Ga)N/GaN microdisk structures have been observed to exhibit room-temperature continuous wave lasing via high-Q whispering gallery modes.2 Work on Si substrates include reports on lasing from GaN microdisks pivoted on Si3 and GaN-based microdisk light-emitting diodes on nanosilicon-on-insulator templates.4 Optically pumped lasing has also been observed in GaN pyramids prepared by lateral epitaxial overgrowth of suitably prepared openings in a Si3N4 mask.5 The performance of all devices fabricated by a top-down approach is, in principle, limited by structural defects present in the as-grown films and by defects generated by the required processing steps. As an alternative, we use a bottom-up approach based on the pendeoepitaxial expansion of self-assembled GaN nanowires (NWs) on Si. This novel and promising approach has been first suggested by Kusakabe et al. for the fabrication of GaN films on foreign substrates based on the coalescence overgrowth of GaN NWs.6 The underlying GaN NW ensembles have been produced either by direct epitaxial growth on the substrate or, more frequently, by etching GaN nanopillars from a GaN film.7 9 In previous work, this technique has been successfully demonstrated for sapphire substrates,7 14 whereas there are only a few reports concerning Si as substrate.15 18 Self-assembled GaN NWs are virtually free of the threading dislocations that form during growth of GaN films and are not r 2011 American Chemical Society

affected by the mismatch in the thermal expansion coefficients between GaN and Si. Consequently, the crystal quality of these GaN NWs is high.19 21 In the present work, we synthesize GaN NWs by molecular beam epitaxy (MBE) on Si(111) and use them as a template for pendeoepitaxy. Choosing growth conditions promoting lateral growth, we demonstrate that the tips of the GaN NWs can be expanded pendeoepitaxially to micrometer size. These GaN microcrystals are found to be entirely free of threading defects even for diameters exceeding 2 μm. Consequently, they exhibit intense and narrow exciton transitions rivaling those observed for state-of-the-art, 500 μm thick freestanding GaN layers prepared by hydride vapor phase epitaxy. GaN NWs were grown on Si(111) substrates by plasmaassisted MBE (PA-MBE) and monitored in situ using reflection high-energy electron diffraction (RHEED). Prior to growth, Si(111) substrates were chemically cleaned by trichloroethylene, acetone, methanol, and a final HF dip (5%, 2 min). After the substrates were loaded into the growth chamber, the substrate surface was covered with several monolayers of Ga and then annealed at 800 °C to remove any residual oxides from the surface. Following the Ga flash-off, the substrate was exposed to atomic N for 5 min for nitridation of the surface prior to NW growth. GaN NWs were subsequently nucleated and grown for 1 h at a substrate temperature of 800 °C and under N-rich conditions (N flux of 15 nm/min, and Ga flux of 5 nm/min). For pendeoepitaxy, after 1 h GaN NW growth, conditions were changed to Ga-stable by keeping the plasma parameters constant, increasing Ga flux to 20 nm/min and decreasing the substrate temperature to 740 °C. The overgrowth was retained for an additional 2 h. The samples were investigated by scanning electron microscopy (SEM), atomic force microscopy (AFM), X-ray diffraction (XRD), transmission electron microscopy (TEM), cathodoluminescence (CL), and microphotoluminescence (μPL) spectroscopy. Top-view and cross-sectional SEM images were recorded with a Hitachi S-4800 at an acceleration voltage of 5 kV. For the Received: June 24, 2011 Revised: August 4, 2011 Published: August 31, 2011 4257

dx.doi.org/10.1021/cg200801x | Cryst. Growth Des. 2011, 11, 4257–4260

Crystal Growth & Design

COMMUNICATION

Figure 1. Cross-sectional SEM image of GaN NWs grown on Si(111) at a substrate temperature of 800 °C for 1 h. The inset displays the RHEED patterns of the GaN NWs during growth recorded along the [1100] and [1120] azimuths.

Figure 3. (a) Top-view SEM image of a detached GaN microcrystal lying upside down on the surface of the coalesced GaN NW ensemble on Si(111) and (b) cross-sectional TEM image of the dislocation-free GaN microcrystal.

Figure 2. (a) Top-view SEM and (b) cross-sectional TEM images of a coalesced GaN NW ensemble on Si(111). The surface of the closed film exhibits GaN microcrystals of a size between 1 to 3 μm embedded in a network of smaller grains. These crystals reveal side facets of {1011}.

cross-sectional TEM analysis using a JEOL JEM-3010 operating at 300 keV, the specimens were prepared by standard mechanical polishing followed by Ar+ milling. μPL measurements were carried out at 10 K using a Cryovac microscope cryostat. The samples were excited with the 325-nm line of a Kimmon He Cd laser focused to a spot size of about 3 μm. CL was performed in a Zeiss Ultra 55 equipped with a Gatan MonoCL 3 system. The typical morphology of the GaN NWs grown on Si(111) is illustrated in Figure 1 by the cross-sectional SEM image. After 1 h of growth, the NWs exhibit an average diameter of about 30 nm and an average length of 0.5 μm. Note that the length distribution

is fairly wide, and some of the NWs are significantly longer than those in their direct vicinity. Furthermore, it is evident that a fraction of the NWs has undergone coalescence with adjacent wires. This unintentional coalescence of the NWs is likely to be caused by the non-negligible tilt of the NWs with respect to the substrate surface.21,22 This tilt is also visible as an arc in the RHEED patterns recorded during growth of this NW sample along the [1100] and [1120] azimuths as shown in the inset of Figure 1. The transmission patterns reveal epitaxial growth with a noticeable orientational distribution. Quantitatively, the tilt of the GaN NWs in Figure 1 is determined to be 3.3° by XRD ω-scans across the GaN(0002) reflection. The μPL spectrum of this GaN NW ensemble (not shown here) is characteristic for highquality GaN NWs in that it is dominated by a narrow donorbound exciton [(D0,XA)] transition at 3.471 eV, that is, at the position corresponding to that of strain-free GaN.23 As a template for pendeoepitaxy, we used a nominally identical GaN NW ensemble with an average length of about 0.5 μm. Figure 2a displays the top-view SEM image of the typical morphology of pendeoepitaxially overgrown and coalesced GaN NWs. A GaN layer on Si(111) of about 1 μm thickness is obtained. The surface of the film reveals the presence of GaN micro- crystals of a size between 1 to 3 μm embedded in a network of smaller grains. The cross-sectional TEM image of the same GaN film in Figure 2b reveals the existence of the {1011} side facets observed at the top of the coalescence boundaries. Furthermore, defects are detected which form at some of the coalescence fronts. 4258

dx.doi.org/10.1021/cg200801x |Cryst. Growth Des. 2011, 11, 4257–4260

Crystal Growth & Design

COMMUNICATION

Figure 5. (a) Top-view SEM and (b) top-view CL images of GaN microcrystals embedded in a network of smaller-sized grains. The emission from the GaN microcrystals is entirely homogeneous and noticeably stronger compared to that of the surrounding network of grains. No threading dislocations are observed.

Figure 4. AFM images of a GaN microcrystal measured over areas of (a) 5  5 μm2 and (b) 2  2 μm2. In (b), atomic steps with a separation of about 200 nm are clearly visible. The height scale is indicated in the figure, and the corresponding rms roughness is 0.2 nm.

Despite the coalescence-induced defects, the top region of the GaN film maintains a very high crystalline quality. Figure 3a depicts the top-view SEM image of a detached GaN microcrystal lying upside down on the surface of the coalesced GaN NW ensemble on Si(111). It is clearly seen that the microcrystal originates from a single GaN NW (white spot in the middle). This is also confirmed by the cross-sectional TEM image of a microcrystal displayed in Figure 3b. Considering that the initial GaN NW array has an inhomogeneous length distribution (cf. Figure 1), we assume that GaN microcrystals are formed by the pendeoepitaxial overgrowth of a single nanowire which is longer than its immediate neighbors. In Figure 3a, the GaN microcrystal is seen to undergo several shape transitions during pendeoepitaxy, until it finally reaches its hexagonal shape. The trigonal symmetry observed at the center of the GaN microcrystal may lead to the formation of rotational domains by creating equivalent surfaces rotated by 180°.24 The existence of the triangular symmetry also explains the asymmetric hexagonalshape observed for some of the GaN microcrystals in Figure 2a (and later in Figure 5a). Most importantly, however, the TEM image reveals that the microcrystal is entirely free of dislocations. AFM performed across one of these microcrystals as shown in Figure 4a reveals a very smooth surface as well as the presence of the {1011} side facets already observed by SEM and TEM. To

obtain the morphology of the microcrystal, we restricted the scan size to an area of 2  2 μm2 as shown in Figure 4b. The microcrystal is seen to exhibit an atomically flat surface with atomic steps separated by 200 nm. The root-mean-square (rms) surface roughness measured over this area amounts to 0.2 nm. The single-crystalline nature of the microcrystals should manifest itself in improved optical properties as well. Figure 5 displays top-view SEM (a) and CL (b) images of a coalesced GaN NW ensemble on Si(111). Three microcrystals surrounded by smaller grains can be recognized. The GaN microcrystals indeed exhibit a significantly stronger emission compared to the surrounding network of small grains. Moreover, the emission of the microcrystals is perfectly homogeneous, reflecting the absence of threading dislocations as already indicated by the TEM image depicted in Figure 3b. The optical quality of the GaN microcrystal is further characterized by μPL and compared to that of a state-of-the-art freestanding GaN layer of 500 μm thickness prepared by hydride vapor phase epitaxy (HVPE) (Figure 6). The spectrum of this free-standing GaN layer is dominated by two distinct peaks that are caused by transitions related to A excitons bound to neutral Si and O donors.25 In addition, a weak acceptor-bound exciton at 3.467 eV and the free A exciton at 3.479 eV are present. The (D0, XA) transition of the GaN microcrystal (which we believe to be related to O rather than to Si) is spectrally located at 3.471 eV revealing strain-free GaN. In comparison, the (D0,XA) transition of the free-standing GaN reference sample is located 0.3 meV higher than 3.471 eV showing that the material is under slight compressive strain. The fwhm of the (D0,XA) transition for the GaN microcrystal amounts to 1 meV, not much larger than that of the free-standing GaN reference sample of 0.6 meV prepared by HVPE. For comparison, the (D0,XA) transition of 4259

dx.doi.org/10.1021/cg200801x |Cryst. Growth Des. 2011, 11, 4257–4260

Crystal Growth & Design

Figure 6. μPL spectrum of a microcrystal under investigation (solid line) compared with that of a state-of-the-art free-standing GaN reference sample prepared by HVPE (shaded area).

the ammonothermally grown GaN (a method of producing bulk GaN crystals) has been reported to be spectrally located 5 meV higher than 3.471 eV (compressive strain) with the narrowest fwhm of 0.46 meV.26 Moreover, the width of 1 meV for the (D0, XA) transition of the GaN microcrystal is much narrower than that of 10 20 meV usually seen for GaN films on Si.17,27,28 To our knowledge, this is the best PL line width obtained for an extended GaN crystal grown by MBE on any substrate. The PL intensity of both samples (Figure 6) is comparable, underlining the high optical quality of the GaN microcrystal obtained by the pendeoepitaxial overgrowth of the underlying GaN NW array. The GaN microcrystals exhibit a structural and optical quality comparable to that of a state-of-the-art free-standing GaN reference sample. To exploit these promising properties, a better control of the GaN microcrystal formation by means of tuning the NW density and length distribution is necessary. Alternatively, patterning of Si substrates would allow us to gain control over the position and size of the individual NWs.29 Moreover, a reduction of the orientational distribution of the NW template is required since the coalescence of misaligned NWs creates a network of boundary dislocations.22 Such a reduction is expected by employing crystalline buffer layers such as AlN30,31 or Si3N4.32

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This work was partially supported by the German BMBF Project GANONSI (Contract No. 13N10255). The authors would like to thank Anne Kathrin Bluhm for SEM images, Claudia Herrmann and Hans-Peter Sch€onherr for technical support, and Rudolf Hey and Raffaella Calarco for valuable discussions. The free-standing GaN layer is courtesy of Ke Xu and Hui Yang from the Suzhou Institute of Nano-Tech and Nano-Bionics. ’ REFERENCES (1) Simeonov, D.; Feltin, E.; Altoukhov, A.; Castiglia, A.; Carlin, J. F.; Butte, R.; Grandjean, N. Appl. Phys. Lett. 2008, 92, 171102. (2) Tamboli, A. C.; Haberer, E. D.; Sharma, R.; Lee, K. H.; Nakamura, S.; Hu, E. L. Nat. Photonics 2007, 1, 61. (3) Choi, H. W.; Hui, K. N.; Lai, P. T.; Chen, P.; Zhang, X. H.; Tripathy, S.; Teng, J. H.; Chua, S. J. Appl. Phys. Lett. 2006, 89, 211101.

COMMUNICATION

(4) Tripathy, S.; Sale, T. E.; Dadgar, A.; Lin, V. K. X.; Zang, K. Y.; Teo, S. L.; Chua, S. J.; Blaesing, J.; Krost, A. J. Appl. Phys. 2008, 104, 053106. (5) Bidnyk, S.; Little, B.; Cho, Y.; Krasinski, J.; Song, J.; Yang, W.; McPherson, S. Appl. Phys. Lett. 1998, 73, 2242. (6) Kusakabe, K.; Kikuchi, A.; Kishino, K. Jpn. J. Appl. Phys. 2001, 40, L192. (7) Tang, T.-Y.; Shiao, W.-Y.; Lin, C.-H.; Shen, K.-C.; Huang, J.-J.; Ting, S.-Y.; Liu, T.-C.; Yang, C. C.; Yao, C.-L.; Yeh, J.-H.; Hsu, T.-C.; Chen, W.-C.; Hsu, H.-C.; Chen, L.-C. J. Appl. Phys. 2009, 105, 023501. (8) Chen, Y.-S.; Shiao, W.-Y.; Tang, T.-Y.; Chang, W.-M.; Liao, C.-H.; Lin, C.-H.; Shen, K.-C.; Yang, C. C.; Hsu, M.-C.; Yeh, J.-H.; Hsu, T.-C. J. Appl. Phys. 2009, 106, 023521. (9) Tang, T.-Y.; Lin, C.-H.; Chen, Y.-S.; Shiao, W.-Y.; Chang, W.-M.; Liao, C.-H.; Shen, K.-C.; Yang, C.-C.; Hsu, M.-C.; Yeh, J.-H.; Hsu, T.-C. IEEE Trans. Electron Devices 2010, 57, 71. (10) Meshi, L.; Cherns, D.; Griffiths, I.; Khongphetsak, S.; Gott, A.; Liu, C.; Denchitcharoen, S.; Shields, P.; Wang, W.; Campion, R.; Novikov, S.; Foxon, T. Phys. Status Solidi C 2008, 5, 1645. (11) Cherns, D.; Meshi, L.; Griffiths, I.; Khongphetsak, S.; Novikov, S. V.; Farley, N.; Cam- pion, R. P.; Foxon, C. T. Appl. Phys. Lett. 2008, 92, 121902. (12) Chao, C. L.; Chiu, C. H.; Lee, Y. J.; Kuo, H. C.; Liu, P.-C.; Tsay, J. D.; Cheng, S. J. Appl. Phys. Lett. 2009, 95, 051905. (13) Li, Q.; Lin, Y.; Creighton, J. R.; Figiel, J. J.; Wang, G. T. Adv. Mater. 2009, 21, 2416. (14) Cherns, D.; Meshi, L.; Griffiths, I.; Khongphetsak, S.; Novikov, S. V.; Campion, R. P.; Foxon, C. T.; Liu, C.; Shields, P.; Wang, W. N. J. Phys.: Conf. Ser. 2010, 209, 012001. (15) Bougrioua, Z.; Gibarta, P.; Calleja, E.; Jahn, U.; Trampert, A.; Ristic, J.; Utrera, M.; Nataf, G. J. Cryst. Growth 2007, 309, 113. (16) Kato, K.; Kishino, K.; Sekiguchi, H.; Kikuchi, A. J. Cryst. Growth 2009, 311, 2956. (17) Yang, T. H.; Ku, J. T.; Chang, J.-R.; Shen, S.-G.; Chen, Y.-C.; Wong, Y. Y.; Chou, W. C.; Chen, C.-Y.; Chang, C.-Y. J. Cryst. Growth 2009, 311, 1997. (18) Ku, J.-T.; Yang, T.-H.; Chang, J.-R.; Wong, Y.-Y.; Chou, W.-C.; Chang, C.-Y.; Chen, C.-Y. Jpn. J. Appl. Phys. 2010, 49, 04DH06. (19) Calleja, E.; Sanchez-García, M. A.; Sanchez, F. J.; Calle, F.; Naranjo, F. B.; Mu~ noz, E.; Jahn, U.; Ploog, K. Phys. Rev. B 2000, 62, 16826. (20) Brandt, O.; Pf€uller, C.; Cheze, C.; Geelhaar, L.; Riechert, H. Phys. Rev. B 2010, 81, 045302. (21) Geelhaar, L. IEEE J. Sel. Top. Quantum Electron. 2011, 17, 878–888. (22) Consonni, V.; Knelangen, M.; Jahn, U.; Trampert, A.; Geelhaar, L.; Riechert, H. Appl. Phys. Lett. 2009, 95, 241910. (23) Kornitzer, K.; Ebner, T.; Grehl, M.; Thonke, K.; Sauer, R.; Kirchner, C.; Schwegler, V.; Kamp, M.; Leszczynski, M.; Grzegory, I.; Porowski, S. Phys. Status Solidi B 1999, 216, 5. (24) Grundmann, M. Phys. Status Solidi B 2011, 248, 805. (25) Wysmolek, A.; Korona, K.; Stepniewski, R.; Baranowski, J.; Bloniarz, J.; Potemski, M.; Jones, R.; Look, D.; Kuhl, J.; Park, S.; Lee, S. Phys. Rev. B 2002, 66, 245317. (26) Bai, J.; Dudley, M.; Raghothamachar, B.; Gouma, P.; Skromme, B.; Chen, L.; Hartlieb, P.; Michaels, E.; Kolis, J. Appl. Phys. Lett. 2004, 84, 3289. (27) Calleja, E.; Sanchez-Garcia, M.; Sanchez, F.; Calle, F.; Naranjo, F.; Munoz, E.; Molina, S.; Sanchez, A.; Pacheco, F.; Garcia, R. J. Cryst. Growth 1999, 201, 296. (28) Dadgar, A.; Poschenrieder, M.; Reiher, A.; Blasing, J.; Christen, J.; Krtschil, A.; Finger, T.; Hempel, T.; Diez, A.; Krost, A. Appl. Phys. Lett. 2003, 82, 28. (29) Schumann, T.; Gotschke, T.; Limbach, F.; Stoica, T.; Calarco, R. Nanotechnology 2011, 22, 095603. (30) Songmuang, R.; Landre, O.; Daudin, B. Appl. Phys. Lett. 2007, 91, 251902. (31) Landre, O.; Bougerol, C.; Renevier, H.; Daudin, B. Nano Technol. 2009, 20, 415602. (32) Wu, C.; Chou, L.; Gwo, S. Appl. Phys. Lett. 2004, 85, 2071. 4260

dx.doi.org/10.1021/cg200801x |Cryst. Growth Des. 2011, 11, 4257–4260